- Convert the input polygon model into a volumetric model
Introduction
As data acquisition technology keeps improving, large polygons models are becomming commonplace. In order to use the massive polygons models, it is often neccassary to obtain a "low resolution" version of the original model. The simplified polygonal models that represent the original model can then be used by a user to interactively explore the dataset or progressively transmitt the model over a medium.A large number of simplication algorithms ( Hoppe, Garland, Rossignac, Varshaney etc.) have been published in the preceding years. However, we believe that these simplification algorithms fail to address two significant issues: topology simplification and processing degenerate polygonal models. For example, Simplification Envelepoes is a technique that offers the tightest error bounds on the ouput that it generates -- the problem is that it expects the input model to be well-behaved ( i.e. a manifold surface ). On the other hand, Vertex Clustering can process any polygonal model and simplify the geometry and topology of the surface. The drawback is that it produces low quality results. This is a conseqeunce of the fact that it uses a regular grid super-imposed on the input model to decide which polygons should be merged with other polgyons. This discretization of space leaves a signature on the output produced by this model -- the output models often contain "blocky" looking surfaces
Our Approach
Our simplification algorithm has 4 stages:- Morpological Operators in the Volumetric domain to simplify the topology
- Isosurface extraction to obtain a polygonal representation of the simplified surface
- Topolgy-preseving triangle count reduction to decimate the isosurface
The
Turbine Blade Model -- shows the four stages of our simplification
pipeline
